Investigation of driver behavior is crucial to predicting roadway capacity and efficiently and safely operating the transportation network. Many studies have investigated and confirmed that imperfect human reactions and heterogeneous behavior impacts traffic flow. Historically, the findings from these studies have been limited in scope due to a lack of high quality trajectory-level driving data needed to develop, calibrate, and validate behavioral models. Despite the recent introduction of new data sources, such as those acquired through naturalistic driving studies (NDS), the scope of research questions have remained minimal due to substantial requirements for data processing. Therefore, to improve the usability of trajectory-level data, this paper contributes two algorithms for systematic data processing. The first algorithm identifies continuous segments with homogeneous driving states. The second algorithm estimates the event that caused a transition between two driving segments. The algorithms were developed using the SHRP2 NDS database; however, the logic is extendable to data collected from other instrumented vehicles. Verification produced 96% accuracy in predicting driving state and 92% accuracy in predicting state transition events. These algorithms are expected to increase the efficiency and reliability of radar data processing to enable research with a broader scope to inform practical decision-making.
Radar-Vision Algorithms to Process the Trajectory-Level Driving Data in the SHRP2 Naturalistic Driving Study
01.11.2018
734078 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Typical Acceleration Profiles for Left-Turn Maneuvers Based on SHRP2 Naturalistic Driving Data
SAE Technical Papers | 2021
|Typical Acceleration Profiles for Left-Turn Maneuvers Based on SHRP2 Naturalistic Driving Data
British Library Conference Proceedings | 2021
|Investigation of Design Speed Characteristics on Freeway Ramps using SHRP2 Naturalistic Driving Data
Transportation Research Record | 2019
|Transportation Research Record | 2021
|